code wiki / _hdl_build / nx_f32_rmsnorm_rope_gate.nx

nx_f32_rmsnorm_rope_gate.nx

buildroot/runtime/_hdl_build/nx_f32_rmsnorm_rope_gate.nx

10370 B133 linesdepth 4pulls 5 transitivereach 0 importersview sourcekind gate/prooftopic f32
docsdependenciesstructsconstsfunctions

dependencies 4 imports · 0 importers

nx_gate_gn.nx nx_gate_base.nx nx_f32_hw.nx nx_syscalls.nx nx_f32_rmsnorm_rope_gate.nx

imports: nx_gate_gn.nxnx_gate_base.nxnx_f32_hw.nxnx_syscalls.nx

imported by: nobody (leaf or entry point)

call flow from main pre-order; caps 40 nodes / depth 6 declared; ↻ = already shown

main gw sys_write f32_div f32_of f32_mul f32_int f32_sin f32_mul ↻ f32_neg f32_div ↻ f32_of ↻ f32_add f32_cos f32_mul ↻ f32_of ↻ f32_neg ↻ f32_div ↻ f32_add ↻ gn sys_mmap rmsnorm_fwd f32_of ↻ f32_add ↻ f32_mul ↻ f32_sqrt f32_of ↻ f32_div ↻ f32_add ↻ f32_div ↻ gm gn ↻ f32_int ↻ f32_mul ↻ f32_of ↻ rmsnorm_bwd sys_mmap ↻ f32_of ↻ f32_mul ↻ f32_add ↻

structs

none

consts

none

functions

18func grow(name: *u8, ok: i64) -> i64 { if ok==1 { gw(" PASS " as *u8) } else { gw(" FAIL " as *u8) } gw(name); gw("
calls 1: gw
20func gm(x: i64) -> i64 { return gn(f32_int(f32_mul(x, f32_of(1000)))) }
called by 1: main
21func f32_abs(x: i64) -> i64 { return x & 0x7FFFFFFF }
called by 1: main calls 1: f32_sub
22func f32_le(x: i64, y: i64) -> i64 { let d: i64=f32_sub(x,y) & 0xFFFFFFFF; if ((d>>31)&1)==1 { return 1 } if (d & 0x7FFFFFFF)==0 { return 1 } return 0 }
called by 1: rmsnorm_fwd calls 3: f32_off32_divf32_add
23func f32_sqrt(x: i64) -> i64 { if (x & 0x7FFFFFFF)==0 { return f32_of(0) } var y: i64=x; var i: i64=0; while i<14 { y=f32_div(f32_add(y, f32_div(x,y)), f32_of(2)); i=i+1 } return y }
25func f32_sin(x: i64) -> i64 { let x2: i64=f32_mul(x,x); var term: i64=x; var sum: i64=x; var k: i64=1; while k<=8 { term=f32_neg(f32_div(f32_mul(term,x2), f32_of((2*k)*(2*k+1)))); sum=f32_add(sum,term); k=k+1 } return sum }
called by 2: rope_Lmain calls 5: f32_mulf32_off32_negf32_divf32_add
26func f32_cos(x: i64) -> i64 { let x2: i64=f32_mul(x,x); var term: i64=f32_of(1); var sum: i64=f32_of(1); var k: i64=1; while k<=8 { term=f32_neg(f32_div(f32_mul(term,x2), f32_of((2*k-1)*(2*k)))); sum=f32_add(sum,term); k=k+1 } return sum }
29func rmsnorm_fwd(x: *i64, gamma: *i64, n: i64, eps: i64, y: *i64, xhat: *i64, rout: *i64) -> i64
36func rmsnorm_bwd(g: *i64, gamma: *i64, xhat: *i64, r: i64, n: i64, dx: *i64) -> i64
43func rms_L(x: *i64, gamma: *i64, w: *i64, n: i64, eps: i64) -> i64
50func rope_L(x0: i64, x1: i64, a: i64, w0: i64, w1: i64) -> i64
56func main() -> i64